AI in Healthcare is getting stronger yet it needs more data governance...
In the recent years, I have heard the argument for and against artificial intelligence (AI) in different industries that its power needs to be tamed. Business leaders such as Elon Musk have raised several concerns over the power of AI where as his own Tesla cars are making good use of technology to deliver self-driving cards. Elon’s concerns should be understood and the power of AI shouldn't harm human interest in anyway shape or form.
Last week Healthcare IT news published an article around “how a mobile app is now able to assess MRI images to analyze dementia? [1]. The app reports by visualizing comparison patient MRI biomarkers with other data patients. This approach is not new to the IT industry, over the last few years augmented reality (AR) has been made more persuasive with its use cases such as being able to take a picture of an object and then the app analyzes the object for its kind, shape, form and other features. Your phone and tablets if equipped with lidar sensor can sense depth very well and measure distances between objects to a greater accuracy. The app for MRI image analysis quantifies and evaluates patient MRI data against the distributions of key dementia-specific imaging biomarkers and reference data from approximately 2,000 patients with a confirmed neurodegenerative disease, including frontotemporal dementia, Alzheimer's disease and vascular dementia reported by the development company. Such applications open up door for more image analysis applications that can help detect disease at very early stages for preventative treatment and accurate time-based diagnosis.
In a report released by ABI Research, AI spending in the health care and pharmaceutical industries is expected to increase from $463 million in 2019 to more than $2 billion over the next five years. [2] Due to COVID19 this figure will surpass $4-5 billion as efforts towards AI have been increased world wide using image detection, temperature measurement and other COVID19 influenced use cases.
Whilst the use of AI becomes more popular and intuitive, the sensitive issue regarding capturing, storing, utilizing and discarding of patient data appropriately has become broad. Companies like Google and Deepmind are facing legal claim for unauthorized use of NHS medical records. The data on 1.6 million patients was allegedly used without their knowledge or consent. DeepMind received the patient data from the Royal Free Hospital (NHS) for the clinical safety testing of a smartphone app called 'Streams', which was developed to detect acute kidney injuries. The UK's data privacy organisation, the Information Commissioner's Office (ICO) subsequently ruled the Royal Free had not complied with the requirements of Data Protection Act when it provided the patient data. [3]
Below are the five checkpoints that should be considered in any AI Healthcare project:
It can be argued that the intentions of AI Healthcare projects is to deliver the best outcome to the patient but rules of the road have to laid out and obeyed by technology providers to supply successful applications which meet the global patient data privacy rules and objectives. More professionals in IT would need to be aware of how to handle patient data as oppose to just handling normal datasets used for AI training and development purposes. In future, the AI apps will be subject to more scrutiny to prove their accuracy levels with the data that they were trained with. Just like we are not 100% ready for self-driving cars as the rules of the present world are complex and dynamic, IT professionals for AI in healthcare will also take time to understand working with patient data is far more complex than working with a predictable subset of data.
References:
DISCLAIMER: All views expressed in this article are solely the opinion of the author and do not represent or relate to any entity that the author has been associated with, is now or will be affiliated in future.
Data governance is a very important thump rule while sharing the data with developers for AI. Very relevant..!
DG and MDM are now a key!
Practical speaking healthcare entities don’t really know where the data is and with more hyped context of AI its getting even more difficult to govern data.